QC across assays & MAE augmentation

library(tidyverse)
library(magrittr)
library(gt)
library(patchwork)
library(SummarizedExperiment)
library(tidySummarizedExperiment)
library(MultiAssayExperiment)

tmp <- fs::dir_map("R scripts/", source)

theme_set(theme_light())

Loading the MultiAssayExperiment object

data_source <- "real"
mae_files <- 
  fs::dir_ls(
    str_c(
      get_data_dir(data_source = data_source),  
      "02 MAEs/"
    ),
    regexp = "mae_full_.*\\.rds$"
  ) |> 
  sort(decreasing = TRUE) |>
  magrittr::extract(1)

mae <- readRDS(mae_files)

rm(mae_files)

Cross-assay QC

Comparing qPCR and metagenomics

tmp <- 
  dplyr::inner_join(
    mae[["mg"]] |> 
      as_tibble() |>
      filter(!is.na(LBP)) |> 
      filter(!is.na(rel_abs_bact)) |> 
      select(.feature, .sample, rel_abs_bact, sample_type) |> 
      dplyr::rename(
        mg_rel_ab_bact = rel_abs_bact
      ),
    mae[["qPCR"]] |>
      as_tibble() |>
      filter(!is.na(LBP)) |> 
      select(.feature, .sample, copies_per_swab_med, copies_per_swab_cv, strain_group_qpcr),
    by = join_by(.feature, .sample)
  )

tmp |> 
  ggplot() +
  aes(x = mg_rel_ab_bact, y = copies_per_swab_med, color = sample_type) +
  geom_point(alpha = 0.3, size = 1) +
  facet_wrap(~ strain_group_qpcr + .feature) +
  scale_y_log10() +
  scale_x_log10()

tmp |> 
  dplyr::count(.feature, mg_rel_ab_bact > 0, copies_per_swab_med > 0) |> 
  ggplot() +
  aes(x = `mg_rel_ab_bact > 0`, y = `copies_per_swab_med > 0`) +
  geom_tile(aes(fill = n)) +
  geom_text(aes(label = n)) +
  scale_fill_gradient(low = "white", high = "dodgerblue3") +
  facet_wrap(~ .feature) 

tmp |> 
  dplyr::count(.feature, mg_rel_ab_bact > 0, copies_per_swab_med > 1e+05) |> 
  ggplot() +
  aes(x = `mg_rel_ab_bact > 0`, y = `copies_per_swab_med > 1e+05`) +
  geom_tile(aes(fill = n)) +
  geom_text(aes(label = n)) +
  scale_fill_gradient(low = "white", high = "dodgerblue3") +
  facet_wrap(~ .feature) 

tmp |> 
  dplyr::count(.feature, mg_rel_ab_bact > 0, copies_per_swab_med > 10000) |> 
  ggplot() +
  aes(x = `mg_rel_ab_bact > 0`, y = `copies_per_swab_med > 10000`) +
  geom_tile(aes(fill = n)) +
  geom_text(aes(label = n)) +
  scale_fill_gradient(low = "white", high = "dodgerblue3") +
  facet_wrap(~ .feature) 

TODO sensitivity-specificity curves for each LBP

Metagenomics and 16S rRNA amplicon sequencing

TODO

Temporal profiles (for 10 “random” participants)

Weekly profiles

weekly_samples_from_randomized_participants <- 
  dplyr::inner_join(
    tibble(uid = mae[["mg"]]$uid),
    tibble(uid = mae[["amplicon"]]$uid),
    by = join_by(uid)
  ) |> 
  dplyr::inner_join(
    mae@colData |> as.data.frame() |> as_tibble() |> 
      filter(randomized) |> 
      select(uid, pid, visit_code, randomized, arm, visit_type),
    by = join_by(uid)
  ) |> 
  filter(visit_type == "Clinic")
  

mae_sub <- mae[, weekly_samples_from_randomized_participants$uid]

random_pids <- mae_sub$pid |> table() |> sort(decreasing = TRUE) |> names() |> head(10)

map(
  random_pids,
  plot_data_for_pid,
  mae_sub = mae_sub
)

[[1]]

[[2]]

[[3]]

[[4]]

[[5]]

[[6]]

[[7]]

[[8]]

[[9]]

[[10]]

Full profiles

all_samples_from_randomized_participants <- 
  dplyr::full_join(
    tibble(uid = mae[["mg"]]$uid),
    tibble(uid = mae[["amplicon"]]$uid),
    by = join_by(uid)
  ) |> 
  dplyr::inner_join(
    mae@colData |> as.data.frame() |> as_tibble() |> 
      filter(randomized) |> 
      select(uid, pid, visit_code, randomized, arm),
    by = join_by(uid)
  )

mae_sub <- mae[, all_samples_from_randomized_participants$uid]

# random_pids <- mae_sub$pid |> table() |> sort(decreasing = TRUE) |> names() |> head(10)

map(
  random_pids,
  plot_data_for_pid,
  mae_sub = mae_sub
)

[[1]]

[[2]]

[[3]]

[[4]]

[[5]]

[[6]]

[[7]]

[[8]]

[[9]]

[[10]]

plot_data_for_pid(selected_pid = "068100046", mae_sub = mae_sub)

We rework this graph for all pids and save it as a png.

plot_path <- str_c(
    get_data_dir(data_source = data_source),  
    "03 QCed MAEs/profiles_plot/")

all_pids <- unique(all_samples_from_randomized_participants$pid)

for (pid in all_pids) {
  p <- plot_data_for_pid(selected_pid = pid, mae_sub = mae_sub)
  
  file_name <- paste0("plot_", pid, ".png")
  
  ggsave(
    filename = str_c(plot_path, file_name),
    plot = p,
    width = 15, height = 8, dpi = 300
  )
}

Participant groups

“Guessing the arms” based on the MG and qPCR data

• “LC-115” if at any visit, two-three LBP115-specific strains are detected >0

• “LC-106-o” if in group 4

• “LC-106-3 or -7” if at any visit, two LBP106-specific strains are detected >0

• otherwise -> placebo

TODO

Primary outcomes

Primary outcome definition

The primary outcome is “colonization with any of the L. crispatus strains contained in the LBP by 5 weeks of follow-up as assessed by metagenomic sequencing of the vaginal microbial community with detection of any one of LBP strains at >5% relative abundance or a combination of the strains accounting for >10% relative abundance by metagenomics.”

By metagenomics

“LBP colonization” at each visit

We compute, at each visit and for each participant, the total relative abundance of all LBP strains or the maximum relative abundance of any LBP strain.

If the total relative abundance of all LBP strains is larger than 0.1 or the maximum relative abundance of any LBP strain is larger than 0.05, we consider that the LBP achieved colonization_mg at that visit.

colonization_LBP_mg <- 
  mae[["mg"]] |> 
  as_tibble() |>
  filter(sample_type == "Clinical sample") |> 
  filter(!is.na(LBP)) |> 
  group_by(.sample) |> 
  summarize(
    n = n(),
    total_LBP_rel_ab_at = sum(rel_abs_bact),
    max_LBP_rel_ab_at = max(rel_abs_bact),
    n_detected_strains_at = sum(rel_abs_bact > 0),
    n_detected_LC106_strains_at = sum(rel_abs_bact[LBP == "LC-106 & LC-115"] > 0),
    n_detected_LC115_only_strains_at = sum(rel_abs_bact[LBP == "LC-115"] > 0)
  ) |> 
  mutate(
    colonized_LBP_at_mg = (total_LBP_rel_ab_at > 0.1) | (max_LBP_rel_ab_at > 0.05)
  ) 

colonization_LBP_mg <-  
  colonization_LBP_mg |> 
  dplyr::left_join(
    mae@colData |> as_tibble() |> 
      select(uid, pid, visit_code, randomized, arm, location) |> 
      dplyr::rename(.sample = uid), 
    by = join_by(.sample)
  ) |> 
  arrange(pid, visit_code) 

colonization_LBP_mg |> 
  ggplot() +
  aes(x = visit_code, y = pid, fill = colonized_LBP_at_mg) +
  geom_tile() +
  facet_grid(
    randomized + arm ~ ., 
    scales = "free_y", space = "free_y", labeller = label_both
    ) +
  theme(    
    strip.text.y = element_text(angle = 0)
  )

“LBP colonization” by each visit

From the colonization_mg status at each visit, we can compute the colonization_mg status by each visit.

A participant is considered to have colonized by a visit if they have been colonized at that visit or any previous visit, starting from their post-product visit (“1200” for all groups).

Since we have some missing visits, we impute these as “not colonized”

colonization_LBP_mg <- 
  colonization_LBP_mg |> 
  select(-c(arm, randomized, location)) |> 
  dplyr::full_join(
    colonization_LBP_mg |> select(pid, arm, randomized, location) |> distinct() |> 
      expand_grid(
        visit_code = unique(colonization_LBP_mg$visit_code) |> sort()
      ),
    by = join_by(pid, visit_code)
  )

colonization_LBP_mg |> 
  ggplot() +
  aes(x = visit_code, y = pid, fill = colonized_LBP_at_mg) +
  geom_tile() +
  facet_grid(
    randomized + arm ~ ., 
    scales = "free_y", space = "free_y", labeller = label_both
    ) +
  theme(
    strip.text.y = element_text(angle = 0)
  )

colonization_LBP_mg <- 
  colonization_LBP_mg |> 
  arrange(pid, visit_code) |>
  mutate(colonized_LBP_at_mg = colonized_LBP_at_mg |> replace_na(FALSE)) |>
  group_by(pid) |> 
  mutate(
    tmp = ifelse(as.numeric(visit_code) < 1200, FALSE, colonized_LBP_at_mg),
    colonized_LBP_by_mg = cummax(tmp) |> as.logical()
  ) |> 
  ungroup() |> 
  select(-tmp)

colonization_LBP_mg |> 
  ggplot() +
  aes(x = visit_code, y = pid, fill = colonized_LBP_by_mg) +
  geom_tile() +
  annotate(
    geom = "rect", 
    xmin = 10.5, xmax = 11.5, ymin = -Inf, ymax = Inf,
    col = "black", alpha = 0, linewidth = 1
  ) +
  facet_grid(
    randomized + arm ~ ., 
    scales = "free_y", space = "free_y", labeller = label_both
  ) +
  theme(
    strip.text.y = element_text(angle = 0)
  )

“Crispatus colonization” at each visit

We compute, at each visit and for each participant, the total relative abundance of all LBP strains or the maximum relative abundance of any LBP strain.

If the total relative abundance of all LBP strains is larger than 0.1 or the maximum relative abundance of any LBP strain is larger than 0.05, we consider that the LBP achieved colonization_mg at that visit.

colonization_crispatus_mg <- 
  mae[["mg"]] |> 
  as_tibble() |>
  filter(sample_type == "Clinical sample") |> 
  filter(species == "crispatus") |> 
  group_by(.sample) |> 
  summarize(
    total_crispatus_rel_ab_at = sum(rel_abs_bact),
    #max_crispatus_rel_ab_at = max(rel_abs_bact),
    #n_detected_crispatus_at = sum(rel_abs_bact > 0)
  ) |> 
  mutate(
    crispatus_dominance_at_mg = (total_crispatus_rel_ab_at > 0.5)
  ) 

colonization_crispatus_mg <-  
  colonization_crispatus_mg |> 
  dplyr::left_join(
    mae@colData |> as_tibble() |> 
      select(uid, pid, visit_code, randomized, arm, location) |> 
      dplyr::rename(.sample = uid), 
    by = join_by(.sample)
  ) |> 
  arrange(pid, visit_code) 

colonization_crispatus_mg |> 
  ggplot() +
  aes(x = visit_code, y = pid, fill = crispatus_dominance_at_mg) +
  geom_tile() +
  facet_grid(
    randomized + arm ~ ., 
    scales = "free_y", space = "free_y", labeller = label_both
    ) +
  theme(    
    strip.text.y = element_text(angle = 0)
  )

“Crispatus colonization” by each visit

From the colonization_mg status at each visit, we can compute the colonization_mg status by each visit.

A participant is considered to have colonized by a visit if they have been colonized at that visit or any previous visit, starting from their post-product visit (“1200” for all groups).

Since we have some missing visits, we impute these as “not colonized”

colonization_crispatus_mg <- 
  colonization_crispatus_mg |> 
  select(-c(arm, randomized, location)) |> 
  dplyr::full_join(
    colonization_crispatus_mg |> select(pid, arm, randomized, location) |> distinct() |> 
      expand_grid(
        visit_code = unique(colonization_crispatus_mg$visit_code) |> sort()
      ),
    by = join_by(pid, visit_code)
  )

colonization_crispatus_mg |> 
  ggplot() +
  aes(x = visit_code, y = pid, fill = crispatus_dominance_at_mg) +
  geom_tile() +
  facet_grid(
    randomized + arm ~ ., 
    scales = "free_y", space = "free_y", labeller = label_both
    ) +
  theme(
    strip.text.y = element_text(angle = 0)
  )

colonization_crispatus_mg <- 
  colonization_crispatus_mg |> 
  arrange(pid, visit_code) |>
  mutate(crispatus_dominance_at_mg = crispatus_dominance_at_mg |> replace_na(FALSE)) |>
  group_by(pid) |> 
  mutate(
    #tmp = ifelse(as.numeric(visit_code) < 1200, FALSE, crispatus_dominance_at_mg),
    crispatus_dominance_by_mg = cummax(crispatus_dominance_at_mg) |> as.logical()
  ) |> 
  ungroup() #|> 
  #select(-tmp)

colonization_crispatus_mg |> 
  ggplot() +
  aes(x = visit_code, y = pid, fill = crispatus_dominance_by_mg) +
  geom_tile() +
  facet_grid(
    randomized + arm ~ ., 
    scales = "free_y", space = "free_y", labeller = label_both
  ) +
  theme(
    strip.text.y = element_text(angle = 0)
  )

LBP colonization VS crispatus dominance

p1 <- colonization_LBP_mg |> 
  ggplot() +
  aes(x = visit_code, y = pid, fill = colonized_LBP_at_mg) +
  geom_tile() +
  facet_grid(
    randomized + arm ~ ., 
    scales = "free_y", space = "free_y", labeller = label_both
    ) +
  theme(
    strip.text.y = element_text(angle = 0)
  )

p2 <- colonization_crispatus_mg |> 
  ggplot() +
  aes(x = visit_code, y = pid, fill = crispatus_dominance_at_mg) +
  geom_tile() +
  facet_grid(
    randomized + arm ~ ., 
    scales = "free_y", space = "free_y", labeller = label_both
    ) +
  theme(
    strip.text.y = element_text(angle = 0)
  )

 p1 + p2 + 
    plot_annotation(
      title = str_c("LBP colonization VS crispatus dominance")
    ) +
    plot_layout(ncol = 2) &
    theme(
      legend.justification = "left"
    )

p1 <- colonization_LBP_mg |> 
  ggplot() +
  aes(x = visit_code, y = pid, fill = colonized_LBP_by_mg) +
  geom_tile() +
  annotate(
    geom = "rect", 
    xmin = 10.5, xmax = 11.5, ymin = -Inf, ymax = Inf,
    col = "black", alpha = 0, linewidth = 1
  ) +
  facet_grid(
    randomized + arm ~ ., 
    scales = "free_y", space = "free_y", labeller = label_both
  ) +
  theme(
    strip.text.y = element_text(angle = 0)
  )

p2 <- colonization_crispatus_mg |> 
  ggplot() +
  aes(x = visit_code, y = pid, fill = crispatus_dominance_by_mg) +
  geom_tile() +
  facet_grid(
    randomized + arm ~ ., 
    scales = "free_y", space = "free_y", labeller = label_both
  ) +
  theme(
    strip.text.y = element_text(angle = 0)
  )

p1 + p2 + 
    plot_annotation(
      title = str_c("LBP colonization VS crispatus dominance")
    ) +
    plot_layout(ncol = 2) &
    theme(
      legend.justification = "left"
    )

By qPCR

“LBP colonization” at each visit

We compute, at each visit and for each participant, the total relative abundance of all LBP strains or the maximum relative abundance of any LBP strain.

If the total relative abundance of all LBP strains is larger than 0.1 or the maximum relative abundance of any LBP strain is larger than 0.05, we consider that the LBP achieved colonization at that visit.

colonization_qPCR <- 
  mae[["qPCR"]] |> 
  as_tibble() |>
  filter(sample_type == "Clinical sample") |> 
  filter(!is.na(LBP)) |> 
  group_by(.sample) |> 
  summarize(
    n = n(),
    total_load_at = sum(copies_per_swab_med),
    n_detected_strains_at = sum(copies_per_swab_med > 0),
    n_detected_LC106_strains_at = sum(copies_per_swab_med[LBP == "LC-106 & LC-115"] > 0),
    n_detected_LC115_only_strains_at = sum(copies_per_swab_med[LBP == "LC-115"] > 0)
  ) |> 
  mutate(
    detected_at_qpcr = total_load_at > 0
  ) 

colonization_qPCR <-  
  colonization_qPCR |> 
  dplyr::left_join(
    mae@colData |> as_tibble() |> 
      select(uid, pid, visit_code, randomized, arm, location) |> 
      dplyr::rename(.sample = uid), 
    by = join_by(.sample)
  ) |> 
  arrange(pid, visit_code) 

colonization_qPCR |> 
  ggplot() +
  aes(x = visit_code, y = pid, fill = detected_at_qpcr) +
  geom_tile() +
  facet_grid(
    randomized + arm ~ ., 
    scales = "free_y", space = "free_y", labeller = label_both
    ) +
  theme(    
    strip.text.y = element_text(angle = 0)
  )

“LBP colonization” by each visit

From the colonization status at each visit, we can compute the colonization status by each visit.

A participant is considered to have colonized by a visit if they have been colonized at that visit or any previous visit, starting from their post-product visit (“1200” for all groups).

Since we have some missing visits, we impute these as “not colonized”

colonization_qPCR <- 
  colonization_qPCR |> 
  select(-c(arm, randomized, location)) |> 
  dplyr::full_join(
    colonization_qPCR |> select(pid, arm, randomized, location) |> distinct() |> 
      expand_grid(
        visit_code = unique(colonization_qPCR$visit_code) |> sort()
      ),
    by = join_by(pid, visit_code)
  )

colonization_qPCR |> 
  ggplot() +
  aes(x = visit_code, y = pid, fill = detected_at_qpcr) +
  geom_tile() +
  facet_grid(
    randomized + arm ~ ., 
    scales = "free_y", space = "free_y", labeller = label_both
    ) +
  theme(
    strip.text.y = element_text(angle = 0)
  )

colonization_qPCR <- 
  colonization_qPCR |> 
  arrange(pid, visit_code) |>
  mutate(colonized_at_qpcr = detected_at_qpcr |> replace_na(FALSE)) |>
  group_by(pid) |> 
  mutate(
    tmp = ifelse(as.numeric(visit_code) < 1200, FALSE, colonized_at_qpcr),
    detected_by_qpcr = cummax(tmp) |> as.logical()
  ) |> 
  ungroup() |> 
  select(-tmp)

colonization_qPCR |> 
  ggplot() +
  aes(x = visit_code, y = pid, fill = detected_by_qpcr) +
  geom_tile() +
  annotate(
    geom = "rect", 
    xmin = 10.5, xmax = 11.5, ymin = -Inf, ymax = Inf,
    col = "black", alpha = 0, linewidth = 1
  ) +
  facet_grid(
    randomized + arm ~ ., 
    scales = "free_y", space = "free_y", labeller = label_both
  ) +
  theme(
    strip.text.y = element_text(angle = 0)
  )

Colonization qPCR VS metagenomics

p1 <- colonization_LBP_mg |> 
  ggplot() +
  aes(x = visit_code, y = pid, fill = colonized_LBP_at_mg) +
  geom_tile() +
  facet_grid(
    randomized + arm ~ ., 
    scales = "free_y", space = "free_y", labeller = label_both
    ) +
  theme(
    strip.text.y = element_text(angle = 0)
  )

p2 <- colonization_qPCR |> 
  ggplot() +
  aes(x = visit_code, y = pid, fill = detected_by_qpcr) +
  geom_tile() +
  facet_grid(
    randomized + arm ~ ., 
    scales = "free_y", space = "free_y", labeller = label_both
    ) +
  theme(
    strip.text.y = element_text(angle = 0)
  )

p1 + p2 + 
    plot_annotation(
      title = str_c("LBP colonization at metagenomics VS qPCR")
    ) +
    plot_layout(ncol = 2) &
    theme(
      legend.justification = "left"
    )

p1 <- colonization_LBP_mg |> 
  ggplot() +
  aes(x = visit_code, y = pid, fill = colonized_LBP_by_mg) +
  geom_tile() +
  annotate(
    geom = "rect", 
    xmin = 10.5, xmax = 11.5, ymin = -Inf, ymax = Inf,
    col = "black", alpha = 0, linewidth = 1
  ) +
  facet_grid(
    randomized + arm ~ ., 
    scales = "free_y", space = "free_y", labeller = label_both
  ) +
  theme(
    strip.text.y = element_text(angle = 0)
  )


p2 <- colonization_qPCR |> 
  ggplot() +
  aes(x = visit_code, y = pid, fill = detected_by_qpcr) +
  geom_tile() +
  annotate(
    geom = "rect", 
    xmin = 10.5, xmax = 11.5, ymin = -Inf, ymax = Inf,
    col = "black", alpha = 0, linewidth = 1
  ) +
  facet_grid(
    randomized + arm ~ ., 
    scales = "free_y", space = "free_y", labeller = label_both
  ) +
  theme(
    strip.text.y = element_text(angle = 0)
  )

p1 + p2 + 
    plot_annotation(
      title = str_c("LBP colonization by metagenomics VS qPCR")
    ) +
    plot_layout(ncol = 2) &
    theme(
      legend.justification = "left"
    )

Adding primary_outcomes assay to MAE

se_primary_col_LBP_mg <-
  SummarizedExperiment(
    assays = list(
      outcome = 
        colonization_LBP_mg |> 
        # filter(!is.na(pid)) |> 
        # mutate(.sample = str_c(pid, "_", visit_code)) |> 
        filter(!is.na(.sample)) |> 
        select(.sample, colonized_LBP_at_mg, colonized_LBP_by_mg) |> 
        as.data.frame() |> 
        column_to_rownames(".sample") |> t()
  ) 
)

se_primary_col_crisp_mg <-
  SummarizedExperiment(
    assays = list(
      outcome = 
        colonization_crispatus_mg |> 
        filter(!is.na(.sample)) |> 
        select(.sample, crispatus_dominance_at_mg, crispatus_dominance_by_mg) |> 
        as.data.frame() |> 
        column_to_rownames(".sample") |> t()
  ) 
)

se_primary_col_LBP_qPCR <-
  SummarizedExperiment(
    assays = list(
      outcome = 
        colonization_qPCR |> 
        filter(!is.na(.sample)) |> 
        select(.sample, detected_at_qpcr, detected_by_qpcr) |> 
        as.data.frame() |> 
        column_to_rownames(".sample") |> t()
    )
  ) 



mae <- c(mae, list(col_LBP_mg = se_primary_col_LBP_mg, 
                   col_crispatus_mg = se_primary_col_crisp_mg,
                   col_LBP_qPCR = se_primary_col_LBP_qPCR))

Saving the MultiAssayExperiment objects

We save the “master” MAE (with all assays); but if needed for publication, we can subset the MAE to only include the assays/data of interest.

saveRDS(
  mae, 
  str_c(
    get_data_dir(data_source = data_source),  
    "03 QCed MAEs/",
    "mae_full_", today() |> str_remove_all("-"), ".rds"
  )
)